Optimal embeddings of ultradistributions into differential algebras (Q1650724): Difference between revisions

The nonlinear theory of generalized functions of J. F. Colombeau solves the problem of embedding distributions into differential algebras while preserving the product of \(C^\infty\)-functions. By a well-known result of \textit{L. Schwartz} [C. R. Acad. Sci., Paris 239, 847--848 (1954; Zbl 0056.10602)], such embeddings are optimal in the sense that preserving the product of \(C^k\)-functions for any finite \(k\) is impossible. A~natural question is whether similar results can be obtained for spaces of ultradistributions and, indeed, this problem has been addressed by a number of researchers over the years. All of these previous attempts, however, have only led to embeddings of \(M_p\)-ultradistributions that preserve the multiplication within some restricted (Denjoy-Carleman) class of ultradifferentiable functions. The paper under review completely resolves this problem. The authors first prove an analogue of the Schwartz impossibility result in the setting of ultradistributions, thereby delineating the maximal possible class of ultradifferentiable functions whose product can be preserved under a Colombeau-type embedding. They then go on to introduce algebras of generalized functions of class \((M_p)\) and \(\{M_p\}\) and study them in some detail. In particular, they derive a projective description of the \(\{M_p\}\)-type algebras and give a point value characterization. They then prove the existence of optimal embeddings in both settings. The last part of the paper is devoted to defining a notion of regular generalized functions analogous to \textit{M. Oberguggenberger}'s space \(\mathcal{G}^\infty\) [Multiplication of distributions and applications to partial differential equations. Harlow: Longman Scientific \& Technical; New York: Wiley (1992; Zbl 0818.46036)], and to lay the foundations of microlocal analysis in this framework by defining corresponding wavefront sets and studying propagation of singularities. Finally, it is shown that the results obtained above can in fact be extended to spaces of ultradistributions defined via weight functions.